diff options
author | Susant Palai <spalai@redhat.com> | 2018-03-16 10:15:52 +0530 |
---|---|---|
committer | Amar Tumballi <amarts@redhat.com> | 2018-04-10 02:39:55 +0000 |
commit | 25b8139c74ae3127905a86dd5f834c39746ad8d7 (patch) | |
tree | 1e400b8ccf17c181b91590e34c22a86043a058dc /xlators/storage/posix/src/posix-helpers.c | |
parent | 13c8bf60155b5ab09d60d4a04f2b0d38014e5fca (diff) |
posix: check file state before continuing with fops
In context of Cloudsync:
In scenarios where a data modification fop e.g. a write landed in
POSIX thinking that the file is local, while the file was actually
remote, can be dangerous. Ofcourse we don’t want to take inodelk
for every read/write operation to check the archival status or
coordinate with an upload or a download of a file. To avoid inodelk,
we will check the status of the file in POSIX it self, before we
resume the fop. This helps us avoiding any races mentioned above.
Now e.g. if a write reached POSIX for a file which was actually remote,
it can check the status of the file and will get to know that the file
is remote. It can error out with this status “remote” and cloudsync
xlator will retry the same operation, once it finished downloading the
file.
This patch includes the setxattr changes to do the post processing of
upload i.e. truncate and setting the remote xattr
"trusted.glusterfs.cs.remote" to indicate the file is REMOTE
Each file will have no xattr if the file is LOCAL, one remote xattr if
the file is REMOTE and a combination of REMOTE and DOWNLOADING xattr if
the file is getting downloaded. There is healing logic of these xattrs
to recover from crash inconsitencies.
Fixes: #387
Change-Id: Ie93c2d41aa8d6a798a39bdbef9d1669f057e5fdb
Signed-off-by: Susant Palai <spalai@redhat.com>
Diffstat (limited to 'xlators/storage/posix/src/posix-helpers.c')
-rw-r--r-- | xlators/storage/posix/src/posix-helpers.c | 482 |
1 files changed, 482 insertions, 0 deletions
diff --git a/xlators/storage/posix/src/posix-helpers.c b/xlators/storage/posix/src/posix-helpers.c index 6d7d8c512db..0ff94df944e 100644 --- a/xlators/storage/posix/src/posix-helpers.c +++ b/xlators/storage/posix/src/posix-helpers.c @@ -2774,3 +2774,485 @@ out: UNLOCK (&fd->inode->lock); return ret; } + +gf_cs_obj_state +posix_cs_heal_state (xlator_t *this, const char *realpath, int *fd, + struct iatt *buf) +{ + gf_boolean_t remote = _gf_false; + gf_boolean_t downloading = _gf_false; + int ret = 0; + gf_cs_obj_state state = GF_CS_ERROR; + size_t xattrsize = 0; + + if (!buf) { + ret = -1; + goto out; + } + + if (fd) { + xattrsize = sys_fgetxattr (*fd, GF_CS_OBJECT_REMOTE, NULL, 0); + if ((xattrsize == -1) && ((errno == ENOATTR) || + (errno == ENODATA))) { + remote = _gf_false; + } else if (xattrsize == -1) { + ret = -1; + gf_msg (this->name, GF_LOG_ERROR, 0, errno, "fgetxattr" + " failed"); + state = GF_CS_ERROR; + goto out; + } else { + remote = _gf_true; + } + + xattrsize = sys_fgetxattr (*fd, GF_CS_OBJECT_DOWNLOADING, NULL, + 0); + if ((xattrsize == -1) && ((errno == ENOATTR) || + (errno == ENODATA))) { + downloading = _gf_false; + } else if (xattrsize == -1) { + ret = -1; + gf_msg (this->name, GF_LOG_ERROR, 0, errno, "fgetxattr" + " failed"); + state = GF_CS_ERROR; + goto out; + } else { + downloading = _gf_true; + } + } else { + xattrsize = sys_lgetxattr (realpath, GF_CS_OBJECT_REMOTE, NULL, + 0); + if ((xattrsize == -1) && ((errno == ENOATTR) || + (errno == ENODATA))) { + remote = _gf_false; + } else if (xattrsize == -1) { + ret = -1; + gf_msg (this->name, GF_LOG_ERROR, 0, errno, "getxattr" + " failed"); + state = GF_CS_ERROR; + goto out; + } else { + remote = _gf_true; + } + + xattrsize = sys_lgetxattr (realpath, GF_CS_OBJECT_DOWNLOADING, + NULL, 0); + if ((xattrsize == -1) && ((errno == ENOATTR) || + (errno == ENODATA))) { + downloading = _gf_false; + } else if (xattrsize == -1) { + ret = -1; + gf_msg (this->name, GF_LOG_ERROR, 0, errno, "getxattr" + " failed"); + state = GF_CS_ERROR; + goto out; + } else { + downloading = _gf_true; + } + } + + if (remote && downloading) { + if (fd) { + ret = sys_fremovexattr (*fd, GF_CS_OBJECT_DOWNLOADING); + } else { + ret = sys_lremovexattr (realpath, GF_CS_OBJECT_DOWNLOADING); + } + + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, errno, + "failed to remove xattr, repair failed"); + state = GF_CS_ERROR; + goto out; + } + + if (buf->ia_size) { + if (fd) { + ret = sys_ftruncate (*fd, 0); + } else { + ret = sys_truncate (realpath, 0); + } + + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, errno, + "truncate failed. File is in inconsistent" + " state"); + state = GF_CS_ERROR; + goto out; + } + } + + state = GF_CS_REMOTE; + goto out; + + } else if (remote) { + if (buf->ia_size) { + if (fd) { + ret = sys_ftruncate (*fd, 0); + } else { + ret = sys_truncate (realpath, 0); + } + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, errno, + "truncate failed. File is in inconsistent" + " state"); + state = GF_CS_ERROR; + goto out; + } + } + + state = GF_CS_REMOTE; + goto out; + } else if (downloading) { + if (buf->ia_size) { + if (fd) { + ret = sys_fremovexattr (*fd, GF_CS_OBJECT_DOWNLOADING); + } else { + ret = sys_lremovexattr (realpath, GF_CS_OBJECT_DOWNLOADING); + } + + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, errno, + "failed to remove xattr, repair failed"); + state = GF_CS_ERROR; + goto out; + } + + state = GF_CS_LOCAL; + goto out; + } + } + + state = GF_CS_LOCAL; +out: + gf_msg_debug (this->name, 0, "heal state returned %d", state); + return state; +} + +gf_cs_obj_state +posix_cs_check_status (xlator_t *this, const char *realpath, int *fd, + struct iatt *buf) +{ + gf_boolean_t remote = _gf_false; + gf_boolean_t downloading = _gf_false; + int ret = 0; + gf_cs_obj_state state = GF_CS_LOCAL; + size_t xattrsize = 0; + int op_errno = 0; + + if (fd) { + xattrsize = sys_fgetxattr (*fd, GF_CS_OBJECT_REMOTE, NULL, 0); + if ((xattrsize == -1) && ((errno == ENOATTR) || + (errno == ENODATA))) { + remote = _gf_false; + } else if (xattrsize == -1) { + ret = -1; + op_errno = errno; + gf_msg (this->name, GF_LOG_ERROR, 0, 0, "getxattr " + "failed err %d", errno); + goto out; + } else { + remote = _gf_true; + } + + xattrsize = sys_fgetxattr (*fd, GF_CS_OBJECT_DOWNLOADING, NULL, + 0); + if ((xattrsize == -1) && ((errno == ENOATTR) || + (errno == ENODATA))) { + downloading = _gf_false; + } else if (xattrsize == -1) { + ret = -1; + op_errno = errno; + gf_msg (this->name, GF_LOG_ERROR, 0, 0, "getxattr " + "failed err : %d", errno); + + goto out; + } else { + downloading = _gf_true; + } + + } + + if (realpath) { + xattrsize = sys_lgetxattr (realpath, GF_CS_OBJECT_REMOTE, NULL, + 0); + if ((xattrsize == -1) && ((errno == ENOATTR) || + (errno == ENODATA))) { + remote = _gf_false; + } else if (xattrsize == -1) { + ret = -1; + op_errno = errno; + gf_msg (this->name, GF_LOG_ERROR, 0, 0, "getxattr " + "failed err : %d", errno); + goto out; + } else { + remote = _gf_true; + } + + xattrsize = sys_lgetxattr (realpath, GF_CS_OBJECT_DOWNLOADING, + NULL, 0); + if ((xattrsize == -1) && ((errno == ENOATTR) || + (errno == ENODATA))) { + downloading = _gf_false; + } else if (xattrsize == -1) { + ret = -1; + op_errno = errno; + gf_msg (this->name, GF_LOG_ERROR, 0, 0, "getxattr " + "failed err : %d", errno); + goto out; + } else { + downloading = _gf_true; + } + } + +out: + if (ret) { + gf_msg ("POSIX", GF_LOG_ERROR, 0, op_errno, "getxattr failed " + "with %d", op_errno); + state = GF_CS_ERROR; + return state; + } + + if ((remote && downloading) || (remote && buf && buf->ia_size)) { + state = GF_CS_REPAIR; + gf_msg_debug (this->name, 0, "status is REPAIR"); + return state; + } + + if (remote) + state = GF_CS_REMOTE; + else if (downloading) + state = GF_CS_DOWNLOADING; + else + state = GF_CS_LOCAL; + + gf_msg_debug (this->name, 0, "state returned is %d", state); + return state; + +} + +int +posix_cs_set_state (xlator_t *this, dict_t **rsp, gf_cs_obj_state state, + char const *path, int *fd) +{ + int ret = 0; + char *value = NULL; + size_t xattrsize = 0; + + if (!(*rsp)) { + *rsp = dict_new (); + if (!(*rsp)) { + gf_msg (this->name, GF_LOG_ERROR, 0, ENOMEM, "failed to" + " create dict"); + ret = -1; + goto out; + } + } + + ret = dict_set_uint64 (*rsp, GF_CS_OBJECT_STATUS, state); + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, ENOMEM, "failed to set " + "dict"); + ret = -1; + goto out; + } + + if (fd) { + xattrsize = sys_fgetxattr (*fd, GF_CS_OBJECT_REMOTE, NULL, 0); + if (xattrsize != -1) { + value = GF_CALLOC (1, xattrsize + 1, gf_posix_mt_char); + if (!value) { + gf_msg (this->name, GF_LOG_ERROR, 0, 0, + "no memory for value"); + ret = -1; + goto out; + } + /* TODO: Add check for ENODATA */ + xattrsize = sys_fgetxattr (*fd, GF_CS_OBJECT_REMOTE, + value, xattrsize + 1); + if (xattrsize == -1) { + gf_msg (this->name, GF_LOG_ERROR, 0, errno, + " getxattr failed for key %s", + GF_CS_OBJECT_REMOTE); + goto out; + } else { + value[xattrsize] = '\0'; + } + } else { + gf_msg (this->name, GF_LOG_ERROR, 0, errno, + " getxattr failed for key %s", + GF_CS_OBJECT_REMOTE); + goto out; + } + } else { + xattrsize = sys_lgetxattr (path, GF_CS_OBJECT_REMOTE, NULL, 0); + if (xattrsize != -1) { + value = GF_CALLOC (1, xattrsize + 1, gf_posix_mt_char); + if (!value) { + ret = -1; + goto out; + } + + xattrsize = sys_lgetxattr (path, GF_CS_OBJECT_REMOTE, + value, xattrsize + 1); + if (xattrsize == -1) { + gf_msg (this->name, GF_LOG_ERROR, 0, errno, + " getxattr failed for key %s", + GF_CS_OBJECT_REMOTE); + goto out; + } else { + value[xattrsize] = '\0'; + } + } else { + gf_msg (this->name, GF_LOG_ERROR, 0, errno, + " getxattr failed for key %s", + GF_CS_OBJECT_REMOTE); + goto out; + } + } + + if (ret == 0) { + ret = dict_set_str (*rsp, GF_CS_OBJECT_REMOTE, value); + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, 0, "failed to set" + "value"); + } + } + +out: + return ret; +} + + +/* This function checks the status of the file and updates the xattr response. + * Also it repairs the state of the file which could have been resulted from a + * crash or transient failures. + */ +int +posix_cs_maintenance (xlator_t *this, fd_t *fd, loc_t *loc, int *pfd, + struct iatt *buf, const char *realpath, dict_t *xattr_req, + dict_t **xattr_rsp, gf_boolean_t ignore_failure) +{ + gf_cs_obj_state state = GF_CS_ERROR; + int ret = 0; + + if (!(dict_get (xattr_req, GF_CS_OBJECT_STATUS) || + dict_get (xattr_req, GF_CS_OBJECT_REPAIR))) + return 0; + + + if (fd) { + LOCK (&fd->inode->lock); + if (dict_get (xattr_req, GF_CS_OBJECT_STATUS)) { + state = posix_cs_check_status (this, NULL, pfd, buf); + gf_msg_debug (this->name, 0, "state : %d", state); + ret = posix_cs_set_state (this, xattr_rsp, + state, NULL, pfd); + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, 0, + "posix_cs_set_state failed"); + } + + if (ignore_failure) { + ret = 0; + goto unlock; + } else { + if (state != GF_CS_LOCAL || ret != 0) { + ret = -1; + goto unlock; + } + } + } + + if (dict_get (xattr_req, GF_CS_OBJECT_REPAIR)) { + state = posix_cs_check_status (this, NULL, pfd, + buf); + gf_msg_debug (this->name, 0, "state : %d", state); + + if (state == GF_CS_REPAIR) { + state = posix_cs_heal_state (this, NULL, + pfd, buf); + + if (state == GF_CS_ERROR) { + gf_msg (this->name, GF_LOG_ERROR, 0, 0, + "repair check failed"); + } + } + + ret = posix_cs_set_state (this, xattr_rsp, + state, NULL, pfd); + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, 0, + "posix_cs_set_state failed"); + if (ignore_failure) + ret = 0; + else + ret = -1; + goto unlock; + } + } + } else { + if (!loc->inode) { + ret = 0; + goto unlock; + } + + LOCK (&loc->inode->lock); + if (dict_get (xattr_req, GF_CS_OBJECT_STATUS)) { + state = posix_cs_check_status (this, realpath, NULL, + buf); + gf_msg_debug (this->name, 0, "state : %d", state); + ret = posix_cs_set_state (this, xattr_rsp, state, + realpath, NULL); + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, 0, + "posix_cs_set_state failed"); + } + + if (ignore_failure) { + ret = 0; + goto unlock; + } else { + if (state != GF_CS_LOCAL || ret != 0) { + ret = -1; + goto unlock; + } + } + } + + if (dict_get (xattr_req, GF_CS_OBJECT_REPAIR)) { + state = posix_cs_check_status (this, realpath, NULL, + buf); + gf_msg_debug (this->name, 0, "state : %d", state); + + if (state == GF_CS_REPAIR) { + state = posix_cs_heal_state (this, realpath, + NULL, buf); + + if (state == GF_CS_ERROR) { + gf_msg (this->name, GF_LOG_ERROR, 0, 0, + "repair check failed"); + } + } + + ret = posix_cs_set_state (this, xattr_rsp, + state, realpath, NULL); + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, 0, + "posix_cs_set_state failed"); + if (ignore_failure) + ret = 0; + else + ret = -1; + goto unlock; + } + } + } + +unlock: + if (fd) + UNLOCK (&fd->inode->lock); + else + UNLOCK (&loc->inode->lock); + + return ret; +} |